The world's first ITU international standard for automotive voice interaction, developed by AISpeech

From April 15 to April 26, 2024, the 16th Study Group (SG16) of the International Telecommunication Union Standardization Bureau (ITU-T) held a plenary meeting in Rennes, France. The framework and requirement for in-vehicle multi-region intelligent speech interaction system, proposed by AISpeech in collaboration with the China Academy of Information and Communications Technology and China Telecom, was discussed at the meeting and officially approved. This is also the first new generation of international standards for automotive voice interaction in this field led by a Chinese company.

About ITU-T SG16

ITU (International Telecommunication Union) is one of the three major international standardization organizations. ITU-T is the ITU Telecommunication Standardization Sector, which is a branch of ITU that specializes in formulating telecommunication standards. ITU-T SG16 is a very active research group of the ITU Standardization Bureau, with the main research areas being multimedia and e-Services. At present, SG16's research scope covers many hot standardization research directions such as video and audio processing and compression coding, video surveillance, content distribution, big data, blockchain, and artificial intelligence.

Gasgoo Automotive Research Institute Data shows that the penetration rate of in-vehicle voice interaction will reach 78.8% in 2023, an increase of 5 percentage points compared to 2022. Among them, the installation rate of new-generation voice interaction-related functions such as wake-up-free, continuous recognition, zoned wake-up, visible and speaking, and voiceprint recognition has increased significantly. Different from the simple voice recognition in the past, the new generation of high-level voice interaction has gradually become the standard of new-era cars.

Source: Gasgoo Automotive Research Institute, "Development Trends of Smart Car HMI Industry"

However, the implementation link of voice interaction technology is relatively long, especially in the smart cockpit environment. It not only includes the full-link connection of voice interaction technology and hardware selection, but also involves the adaptation with the vehicle system, application layer, cloud processing and other links. The current trend of internationalization of the automotive industry chain is getting higher and higher. Different departments of multinational companies, and the R&D teams of automobile companies and technology companies often belong to different countries and regions. The international standards of in-vehicle voice interaction systems are not unified, which may cause adaptation problems in system integration and technology upgrades in the process of cross-regional cooperation, thereby affecting the efficiency and effectiveness of project implementation.

This international standard "Framework and Requirements for In-Vehicle Multi-zone Voice Interaction System" will propose technical capability requirements for the voice input, voice recognition, semantic understanding, voice synthesis, voice output, intelligent decision-making and system docking of the in-vehicle multi-zone voice interaction system. By standardizing the output quality and standards of each link in the high-level voice interaction link in the smart cockpit, the docking efficiency of the technical team will be improved, thereby promoting the improvement of product experience and barrier-free communication of global technology.

As one of the leading units of this international standard, AISpeech has a complete technical innovation system and engineering system R&D capabilities in the full link of conversational artificial intelligence. It has nearly 100 globally original technologies, has authorized nearly 1,500 intellectual property rights, has won many international competitions, and has taken the lead in overcoming the "cocktail party problem" under vehicle cockpit conditions (the most challenging technical problem recognized in the field of speech processing, which refers to: speech separation and recognition and understanding when multiple people speak at the same time), enabling passengers in different seats in the car to interact with the in-vehicle assistant at the same time without affecting each other. This technology has led the transformation of vehicle-mounted intelligent cockpit technology. Up to now, AISpeech has taken the lead/participated in the formulation of nearly 70 national/industry/group standards to promote the standardized development and application of artificial intelligence speech and language technology.

In 2019, AISPEE officially launched its automotive pre-installation business. So far, AISPEE has cooperated with nearly 60 auto brands, including BYD, SAIC-GM-Wuling, Great Wall Motors, BAIC New Energy , Zeekr Auto, Nezha Auto and other top ten Chinese independent brands, with more than 160 mass-produced models and a cumulative "on-vehicle" volume of more than 10 million.

In 2022, AISpeech's automotive voice interaction solution (Tianqin Assistant) passed the China Academy of Information and Communications Technology's trusted AI evaluation and obtained L9 intelligent certification, which is the highest known intelligent level certification for in-vehicle voice interaction products in China.

In 2023, data from Gasgoo Automotive Research Institute's "Development Trends in the Intelligent Vehicle HMI Industry" showed that based on insurance statistics, Sibee's voice-activated car-based systems had the fastest growth rate, reaching 38%.

Source: Gasgoo Automotive Research Institute, "Development Trends of Smart Car HMI Industry"

In 2024, AISpex will upgrade the full-scenario solution application of large-scale model technology and start the overseas strategy. In the automotive field, on the one hand, it will increase localization cooperation with overseas automakers, and on the other hand, it will accompany Chinese new energy automakers to "go global". Recently, the first domestic large-scale automotive model standard compiled by AISpex was officially released ("Large-Scale Pre-trained Model Technology and Application Evaluation Method for Industry Part 4: Automotive" standard), focusing on the high-quality development of the automotive industry and promoting the upgrade and optimization of large-scale automotive model products.

Under the "Belt and Road" initiative, the strategic pace of China's automobile industry going global is accelerating, and the cooperation between Chinese automobile companies, technology companies and overseas partners is becoming closer and closer. The formulation of international standards for in-vehicle multi-zone voice interaction will, on the one hand, fill the gap in this field; on the other hand, it will provide standardized requirements for the technical capabilities and software and hardware interfaces of in-vehicle voice interaction, promote international technical exchanges and cooperation, and help the global development of the automobile industry!